|
Description:
|
We are using the ParameterFactory, "for each property file, invoke one build," to create builds for a UI testing project. Lately (behavior started last week, no change to plugins or Jenkins version), the ParameterFactory does not stop generating sets of builds. After reaching the Parameterized Trigger build step, a correct number of builds are generated (about 72 builds of one project with different parameters), but after an undetermined amount of time between 1 and 10 minutes, another set of 72 builds are generated. This continues in a loop, without stopping, with an observed maximum of over 3000 builds of the same project.
More use case information on the parameterized trigger build step:
- Block until triggered projects finish=false
- Predefined parameters used=true
- Parameters from properties file used=true
- For every property file, invoke one build=true
- File pattern: "test_list/current_run/*.properties"
- Action to perform when no files found: Fail the build step
Logger output for "hudson.plugins.parameterizedtrigger":
Sep 22, 2014 8:55:20 AM INFO hudson.plugins.parameterizedtrigger.FileBuildParameterFactory getParameters
null
Sep 22, 2014 8:55:20 AM INFO hudson.plugins.parameterizedtrigger.FileBuildParameterFactory getParameters
null
Sep 22, 2014 8:55:20 AM INFO hudson.plugins.parameterizedtrigger.FileBuildParameterFactory getParameters
null
Sep 22, 2014 8:55:20 AM INFO hudson.plugins.parameterizedtrigger.FileBuildParameterFactory getParameters
null
Sep 22, 2014 8:55:20 AM INFO hudson.plugins.parameterizedtrigger.FileBuildParameterFactory getParameters
null
Sep 22, 2014 8:55:20 AM INFO hudson.plugins.parameterizedtrigger.FileBuildParameterFactory getParameters
null
(This continues for another ten minutes with the same regularity)
More information:
Rather than blocking the parent project until the test project finishes, we use a system Groovy script to monitor the progress of these tests and enforce a timeout on these projects, in an attempt to curb another problem we experience, project build hangs (the project never passes the "building on [node]" point). Here is the script used:
import hudson.model.*
try {
def env = Thread.currentThread().executable.parent.builds[0].properties.get("envVars");
def projectname = env['CHILD_PROJECT'];
def suite = env['SUITE_NAME'];
def timeout = env['TIMEOUT_IN_MINUTES'].toInteger();
def defaults = [ 'Custom':60, 'Core':45, 'Active':90, 'ActiveFailing':60, 'Random':45, 'All':120, 'Missing':60 ];
if (timeout == 0) { timeout = defaults[suite]; }
def proj = hudson.model.Hudson.instance.getItem(projectname);
println(projectname + " is in the queue: " + proj.isInQueue());
println(projectname + " is building: " + isJobRunning(projectname));
println("Waiting for " + projectname + " to finish with a timeout of " + timeout + " minutes.");
def count = 0;
boolean timedOut = false;
while (isJobRunning(projectname) || proj.isInQueue()) {
count++;
Thread.sleep(60000);
if (count % 10 == 0) {
println(projectname + " is still building and has been building for " + count + " minutes.");
printActiveJobs(runningJobs(projectname), projectname);
}
if (count == timeout) {
println(projectname + " is no longer building. Continuing with job...");
timedOut = true;
break;
}
}
if (!timedOut) { println("All " + projectname + " jobs are done building."); }
else { println("TIMEOUT PERIOD REACHED: " + timeout + " minutes have passed."); }
}
catch (err) {
println("Exception: " + err);
}
def isJobRunning(jobname) {
runningJobs(jobname).find {
if (it.toString().contains(jobname)) {
return true;
}
}
return false;
}
def printActiveJobs(list, name) {
def logentry = "The following " + name +
" builds are still running:\r\n";
list.each {
logentry += it.getDisplayName() + ", ";
}
println(logentry + "\r\n");
}
def runningJobs(jobname) {
def myJobs = hudson.model.Hudson.instance.items.findAll{job ->
job.name==jobname}
|
